Abstract
Introduction: We compared migraine features and acute therapy response in menstrually-related migraines (MRMs) and non-menstrually-related migraines (NMRMs).
Methods: Women with frequent, disabling migraines were prospectively diagnosed with MRM according to the International Classification of Headache Disorders (ICHD-II; N = 107) criteria using a daily electronic headache dairy. Participants received individualized acute therapy while free of prophylactic migraine medications.
Results: Repeated measures logistic regression revealed MRMs were longer (23.4 vs. 16.1 hours, odds ratio [OR] = 1.01, confidence interval [CI] 1.01, 1.02) and more likely associated with disability (85.6% vs. 75.6%, OR = 1.82, CI 1.27, 2.58) than NMRMs. MRMs were also less responsive to acute therapy (two-hour pain-free response = 6.7% vs. 13.4%, OR = .45, CI .26, .80) and reoccurred more frequently within 24 hours after a four-hour pain-free response (36.0% vs. 19.6%, OR = 2.12, CI 1.27, 3.53) than NMRMs.
Discussion: These results support the proposed ICHD-II classification of MRMs and suggest that MRMs may require a treatment approach different from that for NMRMs.
Introduction
Approximately 50% of women with migraine experience menstrually-related migraines (MRMs), or migraines that occur regularly, although not exclusively, during the perimenstrual period (1). Some (2–8) but not other (9–10) studies have found that MRMs are more severe, differ in migraine features, or are less responsive to acute therapy than migraines that occur at other times in the menstrual cycle. Differences in study methodology and patient samples have likely contributed to conflicting findings across studies. Methodological differences include use of retrospective data (3,5,6), lack of prospective MRM diagnosis (2,3,5–9) and inclusion of participants on confounding migraine prophylaxis (2,3,5–9). Subject sample differences include studies using different definitions of MRM (11) and the inclusion of women who may coincidentally experience migraine during the one menstrual cycle studied (2–9) instead of having MRMs in at least two out of three consecutive menstrual cycles as determined by a prospective daily diary, as recommended by the International Classification of Headache Disorders, second edition (ICHD-II)-2 (12). Because women attending headache clinics (2,4,7,8) tend to report that more severe and less treatment-responsive migraines than women from the general population or from primary care settings (3,5,6,9), it has been argued that differences between MRMs and NMRMs are most pronounced in women who experience frequent, severe migraines. Given the current conflicting findings, the distinguishing features, if any, of MRMs and non-menstrually-related migraines (NMRMs) and the validity of a separate classification for MRM remain controversial.
Based on preliminary data, the ICHD-II includes provisional criteria for diagnosis of MRM but confines its definition to the appendix, citing the need for additional data (12). Obtaining these data requires direct comparison of MRMs and NMRMs in women prospectively diagnosed with MRM according to provisional ICHD-II diagnostic criteria. However, to date there have been few of these studies (13,14), and to our knowledge none that also address the methodological limitations that have been noted (e.g. with daily diary assessment of migraine characteristics and/or acute therapy response in women who are not on preventive medication). The present study thus compares migraine characteristics and acute treatment response between MRMs and NMRMs in females prospectively diagnosed with MRMs using provisional ICHD-II criteria (12) who were not taking migraine preventive medication. Data for both the prospective diagnosis of MRM and the assessment of migraine characteristics and treatment response were collected by daily headache diary on a handheld computer.
Methods
Participants and procedures
Participants were recruited for and thus were potential participants in a treatment trial for frequent, disabling migraines (i.e. at least three migraines with disability, with no more than 20 headache days per 30 days) which was unrelated to the topic of MRM. Volunteers were recruited from May 2001 to November 2004 in the communities surrounding Columbus and Athens, Ohio, USA, by self-selection or referral from health care providers. Participants were excluded if they received a diagnosis of probable medication overuse headache or were pregnant, breast-feeding, planning a pregnancy or unwilling to use adequate contraception during the study period. Participants could not use migraine preventive medications (e.g. beta-blockers, anticonvulsants, antidepressants with evidence of effectiveness for migraine) during the assessment of migraine characteristics or acute therapy response. Of 184 women who met the above inclusion/exclusion criteria, 107 met proposed ICHD-II diagnostic criteria for MRM, including migraine without aura occurring in the perimenstrual period (days 0 ± 2 with day 0 the first day of menstrual cycle) in two out of three prospectively recorded menstrual cycles (12). Thirty-eight of these women were taking cyclic birth control medication (e.g. combination oral contraceptives). All participants provided written informed consent approved by the Ohio University Institutional Review Board.
Participants were provided a handheld computer (PalmOS) diary (15) to record information about headache activity (i.e. type of headache [migraine, tension-type, other, unknown], start and end time, date of the headache), severity (not present, mild, moderate or severe), pain and associated symptoms (i.e. nausea, vomiting, phonophobia, photophobia), headache-related disability (i.e. hours of missed unpaid [e.g., housework, childcare, volunteer or schoolwork] and paid work, recreational and social activities and sleep loss due to headache activity), medication usage (i.e. types and amounts of medications taken for the headache) and onset of menstrual flow. All electronic diary data were downloaded into an Access database and reviewed for consistency of diagnostic interpretation and data adequacy by project managers.
Migraine and acute therapy outcome variables were calculated from the electronic headache diary data (15). Menstrually-related migraines were defined as migraines that began on the first day of the menstrual cycle or on two days before and after the onset of menstrual flow (12). Menstruation was defined as endometrial bleeding resulting from either the normal menstrual cycle or from withdrawal of exogenous progestogens, as in the case of combined birth control medication. Pain-free states were defined as a pain intensity score of 0 within the specified time period (16) (e.g. two hours). Associated symptom score was defined as the sum of the nausea (e.g. range = 0, 1, 2, 3), photophobia (e.g. range = 0, 1, 2, 3), phonophobia (e.g. range = 0, 1, 2, 3) and vomiting scores (e.g. range = 0 or 1). Recurrence after pain-free was defined as a return of pain within 24 hours of the initial dosing after the patient has achieved a pain-free state at the specified (e.g. two-hour) time period (15). Disability equivalent hours were defined as hours of disability (unable to perform unpaid or paid work, recreational and social activities plus lost sleep) plus one-half of activity hours with partial (≥50%) disability in functioning (17).
Acute medications were provided for all participants by the treating neurologists according to an individualized protocol. The individualized treatment protocol was based on an assessment of the characteristics of the migraine attacks, associated symptoms, disability and the participant’s previous experience with acute medication. The protocol specified that initial treatment could be a single agent or multiple agents (e.g. triptan, nonsteroidal anti-inflammatory drug [NSAID], anti-emetic) and follow-up treatment if a migraine failed to respond (e.g. additional dose, addition of an NSAID to triptan). Participants were allowed rescue medication (e.g. steroid) if previous acute therapy medications failed to provide relief. The fact that this was a secondary analysis of a data from a study unrelated to MRM minimizes the role of neurologist or participant expectancies or demand characteristics with regard to MRM. The primary acute therapy outcome was the two-hour pain-free response (16). Secondary treatment outcomes included the four-hour pain-free response and recurrence within 24 hours after pain-free at two and four hours after acute migraine treatment.
Statistical methods
To determine the distribution of migraine headaches by day of the menstrual cycle, the percentage of women with a migraine headache on each day of the menstrual cycle was graphically represented. A migraine day was defined as each day that a woman recorded a migraine. Logistic regression was used to estimate the odds ratio (OR) and confidence interval (CI) for women who experienced a migraine day during the interval between two days prior through two days after the first day of the menstrual cycle when compared to all other days. For each headache feature and treatment outcome, the odds of a significant difference between MRMs and NMRMs were determined using logistic regression to estimate ORs and 95% CI. Repeated measures logistic regression (i.e. generalized estimating equations [GEE regression]) was performed to account for intrasubject correlation (1–19). The level of significance was p < .05 for all analyses. Analyses were performed using the SAS system (version 8.2; SAS Institute, Cary, NC) (20). Confidence intervals that include 1.00 are considered nonsignificant. Power was determined post hoc (N = 107) for a two-fold increase in the OR at the 5% level for each measure for repeated measures logistic regression (21) in order to identify tests compromised by inadequate (<70%) power.
Standard protocol approvals, registrations and patients’ consent
Approval for this study was obtained from the Ohio University Institutional Review Board. Written informed consent was obtained from all participants.
Results
Participants
Demographic characteristics of participants with MRMs (N = 107)
GED, graduate equivalency diploma; MRMs, menstrually-related migraines; SD, standard deviation; USD, US dollars.
Of the 107 participants, all met ICHD criteria for the diagnosis of migraine without aura (25). Twenty-three percent of participants were also diagnosed with migraine with aura. In addition, 12 of the 107 participants recorded between 15 and 20 headache days per 30 days, and thus might be classified by ICHD-II as suffering chronic migraine (12). Forty-two percent of participants received a comorbid episodic tension-type headache (TTH) diagnosis, which is consistent with findings for women presenting with disabling intermittent headache in a primary care setting (6).
The participants averaged six migraine attacks (standard deviation [SD] = 2.34), nine migraine days (SD = 4.0), and six disability days (SD = 3.5) per 30 days, with an x2average of one day with TTH (SD = 1.7) per 30 days. Participants recorded a mean of two MRMs (SD = 1.0) and three MRM headache days (SD = 1.9) during acute therapy evaluation (mean = 43.0 days, SD = 11.4). Participants reported experiencing problem headaches for an average duration of 13 years (SD = 8) with four years (SD = 4) of migraines at the current frequency.
Migraine and the menstrual cycle
The 107 participants recorded a total of 208 MRMs and 684 NMRMs. The proportion of participants experiencing a migraine headache (Figure 1) peaked at 55% on the first day of the menstrual cycle (i.e. day 0), with participants almost twice as likely to experience a migraine during the perimenstrual interval of two days prior through two days after the first day of the cycle than at other times during the month (OR = 1.91, CI 1.59, 2.30). No significant association between TTH and the menstrual cycle was evident (OR = 1.02, CI .71, 1.46).
Percentage of participants with migraine or tension-type headache per day of the menstrual cycle; MRM, menstrually-related migraine; HA, headache; MRHA, menstrually-related tension-type headache. MRM headache days = 273; nonmenstrually-related migraine headache days = 922. ****p <.0001, N = 107 clusters. Generalized estimating equation modeling.
Migraine features
Migraine features for menstrually-related migraines and nonmenstrually-related migraines
Note: MRM = Menstrually-related migraine; NMRM = Nonmenstrually-related migraine; Maximum severity = rating of maximum pain (0 = none, 1 = mild, 2 = moderate, 3 = severe); Associated Symptom Score = Sum of phonophobia (0–3), photophobia (0–3), nausea (0–3), and vomiting (0 or 1) scores; Disability Equivalent Hours = Sum of missed activity hours (e.g., unpaid work, paid work, recreational and social activities, lost sleep) plus one half of impaired (<50%) functioning hours.
Medication use
Participants recorded use of acute therapy for 94% (842 of 892) of migraines. Triptans were used in the treatment of 68% (604 of 892) of migraine attacks, alone in 28% (249 of 892) and in combination with analgesics (i.e. NSAIDS) in 40% (355 of 892) of migraine attacks. Analgesics alone were used in the treatment of 27% (238 of 892) of migraine headaches. Rescue medication (e.g. steroid) was used by 24% (26 of 107) of participants to treat only 6.7% (60 out of 892) of migraines. No differences in the proportion of MRMs and NMRMs treated with analgesics, triptans or rescue medication were observed (i.e., all 95% CIs included 1.00). This suggests lack of bias related to treatment protocol, with neither MRMs nor NMRs more likely to have received any single- or multi-agent treatment regimen.
Although the medications used to treat MRMs and NMRMs did not differ, participants used higher doses of triptans and rescue medication to treat MRMs than to treat NMRMs. MRMs were treated with a greater number of triptan doses (2.0) per migraine than NMRMs (1.6; OR = 1.19, 95% CI 1.07, 1.33). Although rescue medication was used infrequently, when taken, MRMs were treated with more doses (2.3) of rescue medication per migraine than were NMRMs (1.7 doses; OR = 1.42, 95% CI 1.08, 1.85). No significant differences (i.e. 95% CI included 1.00) in the number of analgesic pills taken per headache were found between the two types of migraines (MRM = 3.3; NMRM = 3.0 pills).
Acute treatment outcomes
Significant differences in response to acute therapy between MRMs and NMRMs were evident (Figure 2). On the primary outcome measure of two-hour pain-free response, drugs for the acute treatment of migraine were less effective for MRMs than for migraines occurring at other times during the menstrual cycle. Although a two-hour pain-free response was observed in just 12% (105 of 892) of migraines treated with acute therapy, a two-hour pain-free response following acute treatment was observed half as often with MRMs (6.7%; 13 of 195) than with NMRMs (13.4%; 87 of 647; OR = .45, CI .26, .80). However, the four-hour pain-free response observed in 36.2% (87 of 642) of NMRMs and 38.5% (75 of 195) of MRMs (OR = 1.10, 95% CI .81–1.51) did not similarly differentiate MRMs and NMRMs.
Migraine acute treatment outcomes by type of migraine; MRM, menstrually-related migraine; NMRM, nonmenstrually-related migraine; pf, pain-free; 2-hour pf, pain-free 2 hours following acute treatment; 4-hr pf, pain-free 4 hours following acute treatment; Recur after pf @ 4 hr, recurrence within 24 hours after pain-free at 4 hours following acute treatment; Recur after pf @ 2 hr, recurrence within 24 hours after pain-free at 2 hours following acute treatment.
Significant differences in migraine recurrence within 24 hours following acute treatment were observed for MRMs and NMRMs. Recurrence following a four-hour pain-free response to acute therapy was twice as frequent with MRMs than with NMRMs, occurring in 36.0% (27 of 75) of MRMs but only 19.6% (46 of 234) of NMRMs (OR = 2.12, CI 1.27, 3.53). Recurrence following a two-hour pain free-response to acute therapy also was numerically more frequent in MRM, occurring in 38.5% (5 of 13) of MRMs and 14.9% (13 of 87) of NRMs, but this analysis lacked statistical power (<70%).
Discussion
MRMs and NMRMs exhibited marked differences in this sample of women with frequent, severe migraines which met ICHD-II provisional diagnostic criteria for MRMs (12). MRMs were associated with greater disability, were treated with higher doses of acute medication and were less responsive to acute therapy than migraines that occurred at other times during the month. Thus, this research supports the distinction between MRM and NMRM that underlies the provisional ICHD-II classification of MRM (12).
Migraines were almost twice as likely to occur during the perimenstrual period as at other times in the menstrual cycle, as would be expected in women with MRM (1). MRMs were longer in duration than NMRMs, rated slightly more severe and were more frequently accompanied by disability of longer duration. However, in agreement with Granella (4), Stewart (9) and Diamond and colleagues (8), we found no difference in the occurrence of any of the migraine-associated symptoms in MRMs and NMRMs. MRMs were thus qualitatively similar but quantitatively more severe and disabling than NMRMs.
Participants treated MRMs and NMRMs with the same types of acute therapies (e.g. triptans, analgesics). However, MRMs were treated with higher doses of triptans and, when needed, higher doses of rescue medication than were NMRMs. In contrast to the preponderance of the literature, MRMs were less responsive to acute therapy at two hours, and recurred more frequently within 24 hours of a pain-free acute therapy response than did NMRMs. These results are noteworthy because this study adhered to the provisional ICHD-II guidelines in defining MRMs (e.g. prospectively recorded evidence of migraines occurring on days 0 ± 2, with day 0 the first day of menstrual cycle in at least two out of three cycles) and excluded participants receiving concomitant migraine prophylaxis.
Historically, trials of acute therapy with MRM did not restrict their samples to women with prospectively diagnosed MRM (11). Earlier studies typically include all menstruating female migraineurs, with MRMs defined simply as migraines that occur coincidental to the menstrual period during the specified time period examined without requiring the criteria of occurring two out of three consecutive months (8,22,26–29), and recent studies have commonly relied on self-reported history of MRMs (30–35). However, MacGregor and colleagues (36) have shown that retrospective reports of the association of migraines with menstruation are not an accurate substitute for prospective diary recording in the diagnosis of MRM. The inclusion of women not diagnosed with MRMs using prospectively recorded diary data would likely obscure any differences between MRMs and NMRMs (36).
Drugs used for migraine prevention potentially alter metabolism of acute migraine medications or otherwise interact with these medications (16) and, thus, present a potential confound when interpreting research findings. Participant selection criteria often have not specified whether participants are taking migraine preventive medication (29,37) or explicitly stated that preventive migraine medication was permitted (6,26–28,3). The influence of preventive medication on migraine features, disability and response to acute migraine therapy is only beginning to be examined but suggest that preventive mediation is a potential confounding factor for each of these variables (38–40).
This study has a number of limitations. As noted above, subjects were women who experienced frequent, severe migraines. Therefore, the findings may not generalize to women with migraine who experience headaches less frequently. However, we focused on women with frequent, severe migraine because it has been hypothesized that differences between MRMs and NMRMs are most likely to be observed in this subgroup of women (33,41–43). The high level of migraine-associated disability and long history of problem migraines may be one reason for the relatively poor acute treatment response observed in this sample (e.g. only 12% of participants exhibited a two-hour pain-free response after initial acute therapy).
Additionally, acute therapy outcomes in this study were not obtained by a randomized, controlled clinical trial, which is generally considered to be the strongest evidence of the clinical efficacy of therapeutic procedures. However, differences in headache characteristics or acute treatment response between MRMs and NMRMs are best examined within the same woman, using a within-subject or repeated measures analysis as performed in this study, rather than by comparing migraine characteristics or acute treatment response across women within the sample (13).
Finally, the acute therapy protocol permitted acute treatment to vary according to the clinical judgment of the neurologist and the participants’ preferences. While this regimen more closely approximates usual patient care than does the typical clinical trial where all participants use a single medication, we cannot conclude that exclusive use of triptans might not have produced a different treatment response in these women. However, neither MRMs nor NMRMs were more likely to be treated with a triptan or analgesic alone or in combination, and therefore, we did not perform separate analyses for each type of treatment regimen.
Our study found significant differences between MRMs and NMRMs in women with frequent migraines who were not taking preventive medication. We found that MRMs were longer and more likely associated with disability than NMRMs. MRMs also were less responsive to acute therapy and recurred more frequently within 24 hours after a pain-free response. These results suggest that MRMs and NMRMs may require different acute treatment strategies. Currently, acute treatment protocols for MRMs and NMRMs are similar, with the possible exception of greater use of NSAIDs with MRMs than with NMRMs because of the role of prostaglandins in menstrually-related symptoms (44). Multi-agent treatment with a triptan plus NSAIDs (45–49) may improve two-hour treatment response and reduce recurrence when compared to treatment with a triptan alone without a significant increase in adverse effects. Similarly, triptan plus dexamethasone (50) potentially improves two-hour mild pain response relative to triptan alone, but the greater side effects and risks of steroids may outweigh any observed benefits. Alternatively, short-term prophylaxis of MRMs using triptans (50–52) or transcutaneous estradiol (33,53) perimenstrually may prevent MRMs, but additional information about the risks and benefits of long-term triptan prophylaxis is needed. Our findings support further research on such multi-agent treatment strategies and short-term prophylaxis, as well as the evaluation of the effectiveness of preventive medications specifically for MRMs, especially for women experiencing frequent migraine who are diagnosed with MRMs.
Footnotes
Competing interests
Pinkerman declares no competing interests. Holroyd has consulted (on 2 December 2006) and received an investigator-initiated grant from ENDO Pharmaceuticals; he has consulted for Takeda Pharmaceuticals North America (on 12 March 2009). Holroyd has received support from the US National Institutes of Health (NINDS; NS32375). Merck Pharmaceuticals, GlaxoSmithKline Pharmaceuticals and the National Institutes of Health had no role in the data collection, statistical analysis, or writing of this article or the decision to submit for publication.
Acknowledgments
This study was supported by grant NS 32374 from the National Institutes of Health to Kenneth A. Holroyd. We thank Merck Pharmaceuticals and GlaxoSmithKline Pharmaceuticals for their unrestricted donation of triptans. Without the cooperation and assistance from the following people, this research could not have taken place: Connie Cottrell, Francis O’Donnell, Gary Cordingley, Donna Shiels, Jana Drew, Gregg Tkachuk, Linda Kaufman, Sharon Waller, Bernadette Heckman, Carol Nogrady and Kathleen Romenek.